Niche construction mediates climate effects on recovery of tundra heathlands after extreme event

with pellets per year out of 144 plots, and belonged to reindeer (Rangifer tarandus tarandus), small rodents such as Norwegian lemming (Lemmus lemmus), ptarmigan (Lagopus lagopus), grey-sided mountain hare (Lepus timidus) and in the island of Rebbenes (Fig 1) the greylag goose (Anser anser). Experimental design We used a one-factor ANOVA design with extreme event (i.e. removal of all above ground vegetation) as the experimental treatment at each site (Fig 1). Removal of all above ground veg- etation was meant to imitate heavy interacting extreme events such as winter warming epi- sodes [7], burrowing by lemmings/voles [27, 28], or pathogens attacks [29] where large parts (if not all) of the aboveground vegetation dies or might be removed. Because certain types of extreme events can leave a litter layer (for example, a repeated winter warming event will kill the vegetation but not remove it, while voles cut the vegetation at ground level and remove it), this treatment had two sub-treatments: the cut vegetation was left as a litter layer or the cut vegetation was discarded, leaving a bare plot. In June 2009, ten blocks were established in each of the sites (Fig 1). Blocks were placed a minimum of five meters apart to avoid including the same E. nigrum individual in different blocks, since long-lived, clonally reproducing species can have multiple modular units [30]. Unfortunately, two blocks were removed from the study due to unexpected field work prob- lems (one at Skibotn and one at Guhkesja´vri, locations 4 and 5 respectively, Fig 1), leaving 48

blocks and 144 plots in total. Each block included three treatment plots of 50 cm x 50 cm each, i.e. control, extreme event without litter and extreme event with litter (Fig 1). The two extreme event plots were assigned treatments in a random manner and placed beside each other with a 50-cm separation and approximately 2 meters downslope of the control plot (Fig 1). Vegeta- tion was cut with standard garden shears at ground level including the vegetation in a 20-cm buffer zone between plots. Furthermore, all plots (except control plots) were trenched around with a spade which was 30 cm long to eliminate any below-ground effects of the vegetation adjacent to the plot. Environmental measurements Three temperature loggers (Thermochron iButtons1) were placed at ground level at each site to measure soil surface temperature every 3 hours all year round (during 2010–2019). This Table 1. Characteristics of the study sites. Rebbenes Skogsfjord Snarby Skibotn Guhkesja´vri Longitude/Latitude 70˚ 10’; 18˚45’ 69˚ 57’; 19˚ 15’ 69˚ 45’; 19˚ 30’ 69˚ 14’; 20˚ 33’ 69˚ 10’; 20˚ 42’ Elevation (m a.s.l.) 10 264 266 580 510 Mean annual temp. (˚C) 2.2 0.5 0.6 -2 -1.9 Mean annual prec. (mm) 871.18 860.49 917.19 634.16 628.2 Mean summer temp. (˚C, June-July-August) 12.37 11.77 12.04 11.96 11.64 Mean summer prec. (mm, June-July-August) 218.75 179.86 172.16 189.75 202.98 Continentality index 21.8 24.2 23 30.1 31 Freezing days (mean/year) 39 30 5 125 153 Soil pH 4 4 4.5 3.5 4 Soil organic horizon depth (cm) 5–10 3–10 2–5 3–10 5–10

Climatic variables were calculated based on data from the six study years using on-site soil surface temperature loggers, hence temperatures are measured at vegetation level. Precipitation data was gathered from publicly available databases (i.e. senorge.no). https://doi.org/10.1371/journal.pone.0245929.t001 PLOS ONE Niche construction effects on tundra after extreme event PLOS ONE | https://doi.org/10.1371/journal.pone.0245929 February 4, 2021 4 / 14